Absorption of N-Phenylpropenoyl-L-Amino Acids in Healthy Humans by Oral Administration of Cocoa (Theobroma Cacao)

Mol. Nutr. Food Res. 2008, 52, 1201 – 1214 DOI 10.1002/mnfr.200700447 1201

Research Article

Absorption of N-phenylpropenoyl-L-amino acids in healthy humans by oral administration of cocoa (Theobroma cacao)

Timo Stark1, Roman Lang1, Daniela Keller1, Andreas Hensel2 and Thomas Hofmann1

1Lehrstuhl fr Lebensmittelchemie und Molekulare Sensorik, Technische Universitt Mnchen, Freising, Germany 2Institut fr Pharmazeutische Biologie und Phytochemie, Mnster, Germany

Besides flavan-3-ols, a family of N-phenylpropenoyl-L-amino acids (NPAs) has been recently identified as polyphenol/amino acid conjugates in the seeds of Theobroma cacao as well as in a variety of herbal drugs. Stimulated by reports on their biological activity, the purpose of this study was to investigate if these amides are absorbed by healthy volunteers after administration of a cocoa drink. For the first time, 12 NPAs were quantified in human urine by means of a stable isotope dilution analysis with LC-MS/MS (MRM) detection. A maximum amount was found in the urine taken 2 h after the cocoa consumption. The highest absolute amount of NPAs excreted with the urine was found for N- [49-hydroxy-(E)-cinnamoyl]-L-aspartic acid (5), but the highest recovery rate (57.3 and 22.8%), that means the percentage amount of ingested amides excreted with the urine, were determined for N-[49- hydroxy-(E)-cinnamoyl]-L-glutamic acid (6) and N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]-L-tyrosine (13). In order to gain first insights into the NPA metabolism in vivo, urine samples were analyzed by LC-MS/MS before and after b-glucuronidase/sulfatase treatment. As independent of the enzyme treatment the same NPA amounts were found in urine, there is strong evidence that these amides are metabolized neither via their O-glucuronides nor their O-sulfates. In order to screen for caffeic acid O-glucuronides as potential NPA metabolites, urine samples were screened by means of LC-MS/MS for caffeic acid 3-O-b-D-glucuronide and 4-O-b-D-glucuronide. But not even trace amounts of one of these glucuronides were detectable, thus excluding them as major NPA metabolites and underlining the importance of future investigations on a potential O-methylation or reduction of the N-phenylpropenoyl moiety in NPAs.

Keywords: Cocoa / (E)-caffeoyl-3-O-b-D-glucuronide / (E)-caffeoyl-4-O-b-D-glucuronide / N-phenylpropenoyl-L- amino acids / Polyphenols / Received: October 28, 2007; revised: December 30, 2007; accepted: January 13, 2008

1 Introduction nol intake and the risk of cardiovascular disease, indicating various potential flavanol-mediated bioactivities, including Epidemiological and medical anthropological investiga- the improvement of vasodilation [1–3], blood pressure [4], tions suggest that flavanol-rich foods exert cardiovascular insulin resistance, and glucose tolerance [5], the attenuation health benefits. Recent dietary interventions in humans of platelet reactivity [6], as well as the improvement of using polyphenol-containing foods have substantiated epi- immune responses and antioxidant defense systems [7, 8]. demiological data on an inverse relationship between flava- The seeds of Theobroma cacao are an important source of polyphenols comprising 12–18% of its total weight on dry Correspondence: Professor Thomas Hofmann, Lehrstuhl fr Leben- basis. Among the cocoa polyphenols, epicatechin smittelchemie und Molekulare Sensorik, Technische Universitt Mn- (249.4 mg/100 g), and procyanidins such as procyanidin B2 chen, Lise-Meitner-Straße 34, D-85354 Freising, Germany (121.4 mg/100 g), procyanidin C1 (138.6 mg/100 g), and E-mail: [email protected] procyanidin B5 (46.4 mg/100 g) were found as the quantita- Fax: +49-8161-71-2949 tively predominating flavan-3-ols in cocoa nibs used as the Abbreviations: HMBC, heteronuclear multiple bond correlation spec- key ingredient for the manufacturing of chocolate and choc- troscopy; HMQC, heteronuclear multiple-quantum correlation spec- olate confectionary [9]. Besides these flavanoids, a family of troscopy; MRM, multiple reaction monitoring; NPAs, N-phenylprope- N-phenylpropenoyl-L-amino acids 1–10 (Fig. 1) has been noyl-L-amino acids; SIDA, stable isotope dilution assay

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com 1202 T. Stark et al. Mol. Nutr. Food Res. 2008, 52, 1201 – 1214

recently identified as polyphenol/amino acid conjugates in (–)-N-[39,49-dihydroxy-(E)-cinnamoyl]-3-hydroxy-L-tyro- cocoa [10, 11]. Among these compounds, N-[39,49-dihy- sine (3, 28.7 mg/100 g), and N-[49-hydroxy-(E)-cinnamoyl]- droxy-(E)-cinnamoyl]-L-aspartic acid (1, 47.2 mg/100 g) L-aspartic acid (5, 13.9 mg/100 g) were found as predomi-

Figure 1. Chemical structures of N-phenylpropenoyl-L-amino acids 1–13 and their deuterium-labeled

isotopologues d3-1 –d4-13.

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com Mol. Nutr. Food Res. 2008, 52, 1201 – 1214 1203 nating polyphenolic amides in cocoa nibs [9, 12]. Further- chloride, silica gel (70–230 mesh), and formic acid 2 more, amide 3 was identified in red clover (Trifolium pra- (Merck, Darmstadt, Germany). [2,3,5,6- H4]-L-tyrosine tense) [13–15], and together with (–)-N-[49-hydroxy-(E)- was purchased from Cambridge Isotope Laboratories (MA, cinnamoyl]-L-tyrosine (8) was reported as an antioxidant in USA). The enzyme solution containing b-glucuronidase cocoa [16] and as a constituent of the bark of African black- (l100 unit/lL) and sulfatase (l7 unit/lL) from Helix wood (Dalbergia melanoxylon) [17]. In addition (–)-N- pomatia Type HP-2 was purchased from Sigma–Aldrich. [39,49-dihydroxy-(E)-cinnamoyl]-L-tyrosine (4) has been One unit of b-glucuronidase liberated 1.0 lg of phenolph- previously identified in cocoa flowers [18] and together with thalein from phenolphthalein O-glucuronide per hour at the tryptophane derivatives 11 [19] and 12 [20], as well as the pH 5.0 and 378C, one unit of sulfatase hydrolyzed 1.0 lmol tyrosine conjugate 13 were detected in raw robusta coffee of p-nitrocatechol O-sulfate per hour at pH 5.0 and 378C. beans [21]. Very recently, a comprehensive LC-MS/MS Water for chromatographic separations was purified with a screening enabled the identification of several amides in a Milli-Q Gradient A10 system (Millipore, Schwalbach, Ger- variety of herbal drugs; for example, compounds 1, 5, 6, 9, many), and solvents used were of HPLC grade (Merck). and 10 were found in fruits of Coriandrum sativum and flow- Deuterated solvents were obtained from Euriso-Top (Gif- ers of Sambucus nigra as well as Lavandula species, and Sur-Yvette, France). Table water (Evian) from a local sup- compounds 5 and 10 in roots of Angelica archangelica and plier was used for the preparation of the cocoa drink. Alkal- fruits of Phytostigma venenosum [22]. ized cocoa powder prepared from cocoa from Ghana (West Besides their astringent taste activity [9–11] as well as Africa) was obtained from the food industry. their antioxidant properties [16], some of these N-phenylpropenoyl-L-amino acids were found to be biologically 2.2 Preparation of the cocoa drink and analysis of active as phytoallexins [23], as inducers of mitochondrial N-phenylpropenoyl-L-amino acids activity and proliferation rate in human liver cell lines as well as human keratinocytes [22], and as potent inhibitor of A mixture of alkalized cocoa powder (20.3 g) and sucrose the adhesion of Helicobacter pylori to human stomach tis- (3.0 g) was suspended in table water (400 mL) at 508C for sue [22]. In consequence, there is a growing interest in the 10 min whilst stirring. After cooling, this cocoa drink was human bioavailability of these nitrogen-containing phyto- used for the human intervention study as well as for chemi- chemicals from foods. Although it is generally believed that cal analysis. the bioavailability of intact polyphenols including hydroxy- For quantitative determination of N-phenylpropenoyl-L- cinnamic acid esters is rather poor [24], there are no data amino acids by means of a recently developed stable isotope available on the human bioavailability of N-phenylprope- dilution analysis [12], an aliquot of the cocoa drink noyl-L-amino acids consisting of a hydroxycinnamic acid (105.0 g) was spiked with solutions (100 lL, each) of the linked to an amino acid via an amide bond. labeled internal standards d3-1, d3-3, d4-4, d3-5, d3-7, d4-8,

The objective of the present study was, therefore, to d3-10, d5-11, d5-12, and d4-13 in methanol (1.0 mg/mL) and administer a cocoa drink containing defined amounts of N- the mixture was homogenized and equilibrated for 30 min phenylpropenoyl-L-amino acids to healthy volunteers and whilst moving in a lab shaker. Thereafter, the cocoa drink to quantitatively follow the urinary excretion of these was freeze-dried, the powder was extracted with n-pentane amides over time. As the use of stable isotopologues of ana- (5630 mL) for 30 min, and the residual cocoa material was lytes is known to enable the correction of compound dis- extracted five times with acetone/water (70:30 v/v; 30 mL criminating during extraction, cleanup, chromatographic each) for 45 min at room temperature whilst stirring. After separation, as well as MS detection, the recently developed centrifugation, the liquid layer was freed from acetone stable isotope dilution assay (SIDA) with LC-MS/MS mul- under reduced pressure at 308C, and then freeze-dried to tiple reaction monitoring (MRM) detection [12] was used give the acetone/water extract. Aliquots (about 40 mg) of as a versatile and reliable method to ensure the accurate the acetone/water extract were taken up in methanol/water quantification of these phytochemicals. (1:1 v/v; 10 mL) and adjusted to pH 2.5 with formic acid. After membrane filtration, aliquots (5–10 lL) were analyzed by means of LC-MS/MS, which was equipped with a 2 Materials and methods 26150 mm, 5 lm, phenylhexyl column (Phenomenex, Germany) operated with a flow rate of 0.2 mL/min. Chro- 2.1 Chemicals matography was performed starting with a mixture (20:80 The following chemicals and reagents were obtained com- v/v) of methanol and aqueous formic acid (0.1% in water; mercially: caffeic acid, methyl-acetobromo-a-D-glucuro- pH 2.5) for 5 min, then increasing the methanol content to nate, 1,2-dichloroethane, 2,6-di-tert-butyl-4-methylpyri- 60% within 35 min, and finally, to 100% within 5 min. dine, silver trifluoromethane sulfonate, barium hydroxide The labeled compounds d3-1, d3-3, d4-4, d3-5, d3-7, d4-8,

(Sigma–Aldrich, Steinheim, Germany), methanol, sulfuric d3-10, d5-11, d5-12, and d4-13 were used as internal stand- acid, diethyl ether, toluene, ethyl acetate, acetone, sodium ards for the quantitative analysis of the corresponding ana-

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com 1204 T. Stark et al. Mol. Nutr. Food Res. 2008, 52, 1201 – 1214

lytes. In addition, compound 2 was quantified via d3–1, and (20 lL) of the ultrafiltrate obtained was analyzed by means

compounds 6 and 9 via d3-5 as the internal standard. The of LC-MS/MS using the conditions given above for analysis amounts of the individual N-phenylpropenoyl amino acids of the cocoa sample. In order to investigate the liberation of were calculated using the response factors (given in brack- N-phenylpropenoyl-L-amino acids from corresponding ets) for the individual compounds 1 (1.00), 2 (0.80), 3 O-glucuronides and/or O-sulfates, aliquots (4 mL) of the (0.92), 4 (0.94), 5 (0.99), 6 (0.85), 7 (0.86), 8 (0.99), 9 urine samples were adjusted to pH 5.0 with hydrochloric (0.86), 10 (0.85), 11 (0.95), 12 (0.96), 13 (0.90) which had acid (1 mol/L) and, after addition of 10 lL of the b-glucuro- been determined by analysis of solutions containing nidase/sulfatase solution, were incubated for 4 h at 378C defined amounts of the internal standards and the target prior to ultracentrifugation and subsequent LC-MS/MS compounds in five mass ratios from 0.2 to 5.0. analysis.

2.3 Experimental design of human study 2.5 Syntheses

Eight healthy volunteers aging between 24 and 30 years 2.5.1 N-phenylpropenoyl-L-amino acids (four females and four males) participated in the interven- The N-phenylpropenoyl-L-amino acids 1–13 as well as the

tion study. The protocol was fully explained to the volun- isotopologues d3-1, d3-3, d4-4, d3-5, d3-7, d4-8, d3-10, d5-11,

teers who were nonsmokers and gave their written informed and d5-12 (Fig. 1) were synthesized as reported earlier in consent to participate the present study and had no history the literature [11, 12]. Closely following the same proce- of cocoa noncompatibility. Exclusion criteria involved a dure [12], N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]- 2 vegetarian diet, weight-reducing dietary regimen, alcohol- [2,3,5,6- H]-L-tyrosine (d4-13) was synthesized in a yield of ism, consumption of vitamin supplements or nutraceuticals, 40% starting from (E)-ferulic acid and [2,3,5,6-2H]-L-tyro- diabetes, hyperlipidemia, cardiovascular disease, history of sine. gastrointestinal disease, or any other chronic disease. The volunteers followed a controlled diet in which chocolate, 2.5.2 Caffeic acid methylester (14) cocoa, coffee, as well as any coffee or cocoa containing Concentrated sulfuric acid (1 mL) was added to a solution food products were strictly forbidden at least for 48 h prior of caffeic acid (20 mmol) in methanol (100 mL) and the to the intervention. mixture was maintained under nitrogen at room tempera- In the morning, 1 h before the intervention, blank urine ture whilst stirring. After 24 h, diethyl ether (100 mL) and samples were collected in plastic bottles, and 30 min before water (500 mL) were added, the organic layer was sepa- the study, the volunteers consumed a glass of table water rated, and the aqueous phase was extracted with diethyl (200 mL) on an empty stomach. Then, a bolus of the cocoa ether (56100 mL). The combined organic layers were drink, which was freshly prepared as described above, was extracted with water (26200 mL), the organic phase was administered to the eight volunteers. After 1, 2, 3, 4, 6, and freed from solvent in vacuum, and the residue was taken up 8 h, the volunteers collected urine samples which were in methanol (10 mL). Upon addition of water (50 mL), caf- immediately ice-cooled and, then, used for quantitative feic acid methylester (14, yield 91%) was obtained as a analysis. After the first urine collection (after 1 h), the vol- white precipitate, which was isolated by filtration, washed unteers consumed 300 mL of table water, and after the fol- with water (5650 mL), and freeze-dried. lowing urine collections the volunteers consumed 500 mL of table water within 5 min. After the first and the fourth 2.5.3 Caffeic acid O-b-D-glucuronide (15) and 4-O- urine collection, the volunteers had meal together contain- b-D-glucuronide (16) ing rolls, butter, jam, salami, and bacon. Following a modification of the Koenigs-Knorr procedure developed for glucoside synthesis [25], solid caffeic acid methylester (14) (1 mmol), and methyl-acetobromo-a-D- 2.4 Analysis of N-phenylpropenoyl-L-amino acids glucuronate (2 mmol) were added in one portion to a sus- in excreted urine pension of powdered molecular sieve (5 g, 4 ) and silver For the quantitative analysis of N-phenylpropenoyl-L-amino trifluoromethane sulfonate (2 mmol) in anhydrous 1,2- acids in urine, aliquots (5.0 mL) of the urine samples were dichloroethane (30 mL) cooled to –208C with stirring in spiked with solutions (100 lL each) of the labeled internal the dark. After stirring the mixture at –208C for 30 min, a

standards d3-1, d3-3, d4-4, d3-5, d3-7, d4-8, d3-10, d5-11, d5-12, solution of 2,6-di-tert-butyl-4-methylpyridine (2 mmol) in

and d4-13 dissolved in methanol (10 lg/mL) and equili- 1,2-dichloroethane (5 mL) was added, the cooling bath was brated for 30 min at room temperature on a laboratory removed, and the mixture was stirred for additional 14 h at shaker. Aliquots of the samples (500 lL each) were ultracen- r.t. Thereafter, the suspension was filtered, the residue trifuged using a Vivaspin 500 filter cartridge (Satorius washed with 1,2-dichloroethane (2620 mL), silica gel Group, Hannover, Germany) at 12000 rpm by means of a (1 g) was added to the organic solution, and the solvent was Napco 2019R centrifuge for 15 min at 48C. An aliquot carefully evaporated in vacuum. The resulting powder was

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com Mol. Nutr. Food Res. 2008, 52, 1201 – 1214 1205

2 placed onto the top of a glass column (5063cm) filled (400 MHz, DMSO-d6, COSY; the assignment of the NMR- with a slurry of silica gel (5% water) in toluene. Chroma- signals refers to the numbering of the chemical structure tography was performed using toluene, followed by tol- given in Fig. 2): d = 3.67 [s, 36H, H–C(10)], 6.26 [d, uene/ethyl acetate mixtures with stepwise increasing con- 16H, J = 15.66 Hz, H–C(8)], 6.75 [d, 16H, J = 8.34 Hz, tent of ethyl acetate. The fractions eluting with toluene/eth- H–C(5)], 6.99 [dd, 16H, J = 8.08 Hz, J = 2.02 Hz, H– ylacetate (70:30 and 60:40 v/v) were combined, freed from C(6)], 7.04 [d, 16H, J = 2.02, H–C(2)], 7.47 [d, 16H, the solvent in vacuum, and taken up in acetone (50 mL). J = 15.66 Hz, H–C(7)], 9.13 [brs, 16OH, HO–C(3)], 9.58 13 After ice-cooling to 08C, an aqueous Ba(OH)2 solution [brs, 16OH, HO–C(4)]; C-NMR (100 MHz, DMSO-d6,

(0.12 mol/L, 250 mL) was added and the mixture was HMBC, HMQC): d = 51.3 [CH3, C(10)], 113.3 [CH, C(8)], shaken for 10 min. After maintaing the mixture at 08C for 114.7 [CH, C(2)], 115.3 [CH, C(5)], 121.4 [CH, C(6)],

1 h, another aliquot (150 mL) of an aqueous Ba(OH)2 solu- 125.4 [C, C(1)], 144.9 [CH, C(7)], 145.4 [HO–C, C(3)], tion (0.12 mol/L) was added and the mixture was main- 148.4 [HO–C, C(4)], 167.0 [COO, C(9)]. tained for 12 h in the dark under an atmosphere of argon at 08C. Then, the solution was acidified with aqueous sulfuric 2.6.3 Caffeic acid 3-O-b-D-glucuronide (15) acid (1 mM), precipitated BaSO4 was removed by filtration, UV/Vis (water/MeOH, 28:72 v/v; 0.1% formic acid): – the filtrate was concentrated to about 10 mL in vacuum, kmax = 238, 292, 317 nm; LC/MS (ESI) : m/z (%) 355 (100, and aliquots (1 mL) were separated by means of HPLC on [M–H] – ); MS/MS (ESI – ): m/z 178.9 (100%), 135 (47%), RP18 (Varian, Microsorb 100-5, 250621.2 mm2,5lm) 175.0 (23%), 113 (23%); exact mass: m/z 379.0637+ (calcd. + 1 using acidified water (0.1% formic acid) as eluent A and for C15H16O10Na : 379.0636); H-NMR (400 MHz, DMSO- methanol (0.1% formic acid) as eluent at a flow rate of d6, COSY; the assignment of the NMR-signals refers to the 20 mL/min. After isocratic elution with 95% eluent A for numbering of the chemical structure given in Fig. 2): 5 min, the content of eluent B was increased to 10% within d = 3.33 [m, 26H, H–C(29), H–C(39)], 3.40 [pt, 1H, 5 min, then to 20% within 10 min, to 25% within 13 min, J = 9.2 Hz, H–C(49)], 3.98 (d, 1H, J = 9.6 Hz, H–C(59)), and finally, to 100% within 2 min, followed by isocratic 4.49 [brs, 1OH, HO–C(49)], 5.07 [d, 1H, J = 7.2 Hz, H– elution for 5 min. Monitoring the effluent at 324 nm, the C(19)], 5.22 [brs, 1OH, HO–C(29) or HO–C(39)], 5.47 [brs, fractions containing caffeic acid 3-O-b-D-glucuronid (15, 1OH, HO–C(29), or HO–C(39)], 6.31 [d, 1H, J = 15.9 Hz, yield: l10%) and 4-O-b-D-glucuronid (16, yield, l6%) H–C(8)], 6.84 [d, 1H, J = 8.4 Hz, H–C(5)], 7.19 [dd, 1H, were collected individually, freed from solvent in vacuum, J = 8.4 Hz, J = 1.8 Hz, H–C(6)], 7.36 [d, 1H, J = 1.8 Hz, and freeze-dried for 48 h. H–C(2)], 7.42 [d, 1H, J = 15.9 Hz, H–C(7)], 9.30 [brs, 1OH, HO–C(4)], 12.25 [brs, 2OH, HOOC(9), HOOC(69)]; 13C-NMR (100 MHz, DMSO-d , HMBC, HMQC, DEPT- 2.6 Spectroscopic data 6 135): d = 71.3 [CH, C(49)], 72.9 [CH, C(29)], 75.2 [CH, 2.6.1 N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]- C(39)], 75.3 [CH, C(59)], 100.5 [CH, C(19)], 115.0 [CH, 2 [2,3,5,6- H]-L-tyrosine (d4-13) C(2)], 115.9 [CH, C(8)], 116.2 [CH, C(5)], 124.1 [CH,

UV/Vis (MeOH/water, 1:1 v/v; pH 2.5) kmax = 221, 297, C(6)], 125.7 [C, C(1)], 144.1 [CH, C(7)], 145.0 [C, C(3)], 322 nm; LC/MS (ESI+): m/z 362 (100%, [M + 1]+), 384 149.0 [C, C(4)], 167.8 [COO, C(9)], 170.1 [COO, C(69)]. (55%, [M + 23]+), 177 (30%, [M – 184]+); MS/MS (ESI+): 177 (100%, [M–184]+), 89 (76%, [M–272]+), 145 (70%, 2.6.4 Caffeic acid 4-O-b-D-glucuronide (16) + + + [M–216] ), 117 (51%, [M–244] ), 362 (1%, [M + 1] ); UV-Vis (water/MeOH, 37:63 v/v; 0.1% formic acid): kmax + + – exact mass: m/z 384.1352 (Calcd. for C19H15D4NO6Na = 240, 289, 321 nm; LC/MS (ESI) : m/z 355 (100%, [M– + 1 – – 384.1356 ); H NMR (400 MHz, DMSO-d6, COSY; the H] ); MS/MS (ESI ): m/z 179.0 (100%), 135 (47%), 175.0 assignment of the NMR-signals refers to the numbering of (24%) 113 (23%); exact mass: m/z 379.0637+ (Calcd. for + 1 the chemical structure as displayed in Fig. 1): d = 2.80 [dd, C15H16O10Na : 379.0636); H-NMR (400 MHz, DMSO-d6, 1H, J = 9.2, 14.0 Hz, H–C(7a)], 2.98 [dd, 1H, J = 4.6, COSY; the assignment of the NMR-signals refers to the 14.0 Hz, H–C(7b)], 3.81 [s, 3H, H–C(109)], 4.46 [ddd, 1H, numbering of the chemical structure given in Fig. 2): J = 4.6, 8.8, 14.0 Hz, H–C(8)], 6.54 [d, 1H, J = 15.6 Hz, d = 3.33 [m, 26H, H–C(29), H–C(39)], 3.38 [m, 1 H, H– H–C(89)], 6.79 [d, 1H, J = 8.0 Hz, H–C(59)], 6.98 [dd, 1H, C(49)], 3.82 [d, 1H, J = 9.2 Hz, H–C(59)], 4.93 [d, 1H, J = 1.6, 8.0 Hz, H–C(69)], 7.12 [d, 1H, J = 1.6 Hz, H– J = 7.2 Hz, H–C(19)], 5.19 [brs, 1OH, HO–C(29)orHO– C(29)], 7.28 [d, 1H, J = 15.6 Hz, H–C(79)], 8.15 [d, 1H, C(39)], 5.46 [brs, 1OH, HO–C(29) or HO–C(39)], 5.78 [brs, J = 8.0 Hz, H–N], 9.21 [brs, 1OH, HO–C(4)], 9.46 [brs, 1OH, HO–C(49)], 6.31 [d, 1H, J = 16.0 Hz, H–C(8)], 7.05 1OH, HO–C(49)], 12.72 [brs, 1OH, HOO–C(9)]. [d, 1H, J = 8.4 Hz, H–C(5)], 7.09 [dd, 1H, J = 8.4 Hz, J = 1.6 Hz, H–C(6)], 7.13 [d, 1H, J = 1.6 Hz, H–C(2)], 2.6.2 Caffeic acid methylester (14) 7.45 [d, 1H, J = 1.6 Hz, H–C(7)], 8.90 [brs, 1H, HO– 13 UV-Vis (MeOH, 0.1% formic acid): kmax = 294, 329 nm; C(3)], 12.25 [brs, 2OH, HOOC(9), HOOC(69)]; C-NMR – – 1 LC/MS (ESI ): m/z 193 (100%, [M–H] ); H-NMR (100 MHz, DMSO-d6, HMBC, HMQC, DEPT-135):

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com 1206 T. Stark et al. Mol. Nutr. Food Res. 2008, 52, 1201 – 1214

Figure 2. Synthetic sequence used for the preparation of caffeoyl-3-O-b-D-glucuronide (15) and caffeoyl-4-O-b-D-glucuronide (16).

d = 71.5 [CH, C(49)], 72.9 [CH, C(29)], 75.1 [CH, C(39)], The HPLC-MS/MS parameters used were the same as 75.2 [CH, C(59)], 101.0 [CH, C(19)], 115.0 [CH, C(2)], applied for the analysis of caffeic acid O-b-D-glucuronides 116.0 [CH, C(5)], 117.2 [CH, C(8)], 120.4 [CH, C(6)], in urine samples. 128.9 [C, C(1)], 143.8 [CH, C(7)], 146.8 [C, C(3)], 146.9 [C, C(4)], 167.6 [COO, C(9)], 170.5 [COO, C(69)]. 2.8 HPLC The HPLC apparatus (Kontron, Eching, Germany) con- 2.7 Enzymatic hydrolysis of caffeic acid O-b-D- sisted of low pressure gradient system 525 HPLC pump, an glucuronides M800 gradient mixer, a type 560 autosampler, and a DAD A solution of the caffeic acid O-b-D-glucuronides 15 or 16 type 540+ diode array detector. Chromatography was per- (1 mg, each) in water (10 mL) was adjusted to pH 5.0 with formed on 25064.6 mm stainless-steel columns packed aqueous hydrochloric acid (1 mol/L), spiked with 10 lLof with RP-phenylhexyl material, 5 lm (Phenomenex) oper- the b-glucuronidase/sulfatase solution, and then, incubated ated with a flow rate of 0.8 mL/min. for 6 h at 378C. After 0.5, 1, 2, 3, 4, and 6 h, respectively the mixtures were ice-cooled and aliquots (5 lL) were immedi- 2.9 LC-MS/MS ately analyzed by means of HPLC-MS/MS equipped with a 15062mmid,5lm, RP-phenylhexyl column (Phenom- LC-MS/MS analysis was performed using an Agilent 1100 enex) operated with a flow rate of 0.2 mL/min. Chromatog- HPLC-system connected to an API3200-type LC-MS/MS raphy was performed starting with a mixture (20:80 v/v) of system (Applied Biosystems, Darmstadt, Germany) run- methanol and aqueous formic acid (0.1% in water, pH 2.5), ning in the ESI mode. For ESI+, the ion spray voltage was then the methanol content was increased to 100% within set at +5500 V in the positive mode and for ESI – at 25 min, and, finally, held at 100% for 5 min. Using the –4500 V in the negative mode. Zero grade air served as neb- MRM with ESI – ionization, the individual caffeic acid ulizer gas (35 psi), and as turbo gas (3008C) for solvent dry- O-b-D-glucuronides 15 and 16 were analyzed using the ing (45 psi). Nitrogen served as curtain (20 psi) and colli- mass transitions m/z 355.2 fi 179.0, m/z 355.2 fi 135.0, sion gas (4.5610 –5Torr). Both quadrupols were set at unit and m/z 355.2 fi 113.0 monitored for a duration of 250 ms. resolution. By means of the MRM mode, the individual

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com Mol. Nutr. Food Res. 2008, 52, 1201 – 1214 1207

N-phenylpropenoyl amino acids were analyzed using the Table 1. Concentrations of N-phenylpropenoyl-L-amino acids following mass transitions monitored for a duration of in the cocoa drinka) used for the intervention study 75 ms as given in brackets: 1 (m/z 296.3 fi 163.0), d -1 (m/ 3 Compound (no.) Conc. z 299.3 fi 163.0), 2 (m/z 310.3 fi 163.0), 3 (m/z 360.3 fi (lg/100g)b) 163.0), d3-3 (m/z 363.3 fi 163.0), 4 (m/z 344.3 fi 163.0), d4-4 (m/z 348.3 fi 163.0), 5 (m/z 280.3 fi 147.0), d3-5 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-aspartic acid (1) 1196.2 (m/z 283.3 fi 147.0), 6 (m/z 294.3 fi 147.0), 7 (m/z 344.3 N-[49-hydroxy-(E)-cinnamoyl]-L-aspartic acid (5) 473.3 N-[49-hydroxy-(E)-cinnamoyl]-L-tyrosine (8) 195.7 fi 147.0), d -7 (m/z 347.3 fi 147.0), 8 (m/z 328.3 fi 3 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-dopa (3) 106.2 147.0), d4-8 (m/z 332.3 fi 147.0), 9 (m/z 310.3 fi 177.0), N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]-L-aspartic acid (9) 61.4 10 (m/z 264.3 fi 131.0), d3-10 (m/z 267.3 fi 131.0), 11 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-tyrosine (4) 39.8 N-[49-hydroxy-(E)-cinnamoyl]-L-dopa (7) 26.1 (m/z 367.3 fi 163.0), d5-11 (m/z 72.3 fi 163.0), 12 N-[49-hydroxy-(E)-cinnamoyl]-L-glutamic acid (6) 18.4 (m/z 351.3 fi 147.0), d5-12 (m/z 356.3 fi 147.0) 13 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-glutamic acid (2) 16.0 (m/z 358.3 fi 177.0), and d4-13 (m/z 362.3 fi 177.0). N-[cinnamoyl]-L-aspartic acid (10) 10.6 N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]-L-tyrosine (13) 3.7 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-tryptophane (11) 1.9 2.10 LC/TOF-MS N-[49-hydroxy-(E)-cinnamoyl]-L-tryptophane (12) 1.7

Mass spectra of the target compounds were measured on a a) A mixture of alkalized cocoa powder (20.3 g) and sucrose Bruker Micro-TOF (Bruker Daltronics, Bremen, Germany) (3.0 g) was suspended in table water (400 mL) at 508C for mass spectrometer with flow injection referenced on 10 min. After cooling, this drink was used for quantitative sodium formiate and PEG 600, respectively. The com- analysis of amides. pounds were dissolved in 1 mL of MeOH and 10 lLofa b) Concentrations are given as the mean of three independent clean-ups and measurements of the sample. saturated solution of NaBF4 in MeOH was added to meas- ure the exact mass of the sodium adducts. (E)-cinnamoyl]-L-tyrosine (8) and N-[39,49-dihydroxy-(E)- cinnamoyl]-3-hydroxy-L-tyrosine (3) with somewhat lower 2.11 NMR amounts of 195.7 and 106.2 lg per 100 g of cocoa drink, 1H, COSY, heteronuclear multiple-quantum correlation respectively (Table 1). All the other N-phenylpropenoyl-L- spectroscopy (HMQC), heteronuclear multiple bond corre- amino acids were present in amounts below 61.4 lg per lation spectroscopy (HMBC), 13C, and DEPT-135 NMR 100 g of cocoa drink. measurements were performed on a DMX 400 spectrometer In order to gain first insights into the absorption of these (Bruker, Rheinstetten, Germany). Chemical shifts were ref- polyphenolic amides upon cocoa consumption, a bolus of erenced to the solvent signal. Data processing was perform- the freshly prepared cocoa drink was administered to eight ed by using XWin-NMR software (version 3.5; Bruker) as healthy volunteers who followed a controlled diet lacking well as Mestre-C software (Mestrelab Research, Santiago any chocolate, cocoa, coffee, as well as any coffee or cocoa de Compostela, Spain). containing food products for at least 48 h prior to the intervention. Urine samples were collected in plastic bottles 1 h before starting the study, and 1, 2, 3, 4, 6, and 8 h after 3 Results ingestion of the cocoa drink. After cooling, the urine sam-

ples were spiked with defined amounts of d3-1, d3-3, d4-4, 3.1 Quantification of N-phenylpropenoyl-L-amino d3-5, d3-7, d4-8, d3-10, d5-11, d5-12, and d4-13 as the internal acids in administered cocoa drink and urine standards, followed by equilibration, and clean-up by ultra- samples centrifugation. Analysis of the N-phenylpropenoyl-L-amino For the accurate quantitative analysis of the N-phenylprope- acids was performed by means of analytical RP-HPLC-MS/ noyl-L-amino acids in the cocoa drink to be used for the MS (MRM) as exemplified for compounds 1, 5, and 8 in intervention study, a freshly prepared cocoa drink was Fig. 3. As given in Table 2, the blank urine taken 1 h prior to spiked with defined amounts of d3-1, d3-3, d4-4, d3-5, d3-7, the intervention did not contain any detectable amounts of d4-8, d3-10, d5-11, d5-12, and d4-13 as the internal standards, N-phenylpropenoyl-L-amino acids. Independent from the followed by homogenization and equilibration at room tem- chemical structure of the N-phenylpropenoyl-L-amino acid perature. After de-fatting and acetone/water extraction of analyzed, a maximum amount was found in the urine sam- the N-phenylpropenoyl-L-amino acids, mass chromatogra- ples taken 2 h after the cocoa consumption. Thereafter, the phy was performed by means of analytical RP-HPLC-MS/ amount of N-phenylpropenoyl-L-amino acid was found to MS. By far the highest amounts of 1196.2 and 473.3 lg per decrease again with increasing time, e.g., after 8 h the target 100 g of cocoa drink were found for N-[39,49-dihydroxy- amides were not detectable or present just in trace amounts, (E)-cinnamoyl]-L-aspartic acid (1) and N-[49-hydroxy-(E)- respectively. Among all urine samples analyzed, by far the cinnamoyl]-L-aspartic acid (5), followed by N-[49-hydroxy- highest amount of 64.37 l 20.22 lg was measured for

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com 1208 T. Stark et al. Mol. Nutr. Food Res. 2008, 52, 1201 – 1214

Figure 3. RP-HPLC-MS/ MS(ESI+)-MRM analysis for the quantitative determination of the N-phenylpropenoyl-L-amino acids 1, 5, and 8 in urine excreted after 1 h using the internal standards

d3-1, d3-5, and d4-8.

Table 2. Amount of N-phenylpropenoyl-L-amino acids excreted with urine over a time period of 8 h after intervention

Compound no. Amounta) (lg) excreted after

0h 1h 2h 3h 4h 6h 8h

1 n.d. 3.59 l 3.55 4.76 l 4.01 3.64 l 2.99 1.74 l 1.80 0.83 l 1.17 0.48 l 1.05 2 n.d. n.d. n.d. n.d. n.d. n.d. n.d. 3 n.d. 0.52 l 1.24 1.25 l 1.19 <0.01 <0.01 n.d. n.d. 4 n.d. 0.04 l 0.12 0.30 l 0.39 <0.01 <0.01 n.d. n.d. 5 n.d. 22.92 l 8.81 64.37 l 20.22 31.96 l 7.04 16.54 l 5.69 11.10 l 2.44 3.01 l 1.93 6 n.d. 8.98 l 4.60 22.17 l 12.57 7.21 l 2.79 3.22 l 1.55 2.06 l 1.54 0.92 l 1.18 7 n.d. 0.26 l 0.51 0.85 l 0.96 a0.01 a0.01 n.d. n.d. 8 n.d. 10.86 l 5.48 18.91 l 5.53 7.44 l 1.44 3.56 l 1.30 2.53 l 0.66 0.73 l 0.48 9 n.d. 4.67 l 1.66 9.70 l 2.93 4.11 l 0.95 2.07 l 0.57 1.02 l 0.92 0.08 l 0.21 10 n.d. 0.06 l 0.18 0.43 l 0.33 0.14 l 0.20 0.07 l 0.14 a0.01 n.d. 11 n.d. n.d. n.d. n.d. n.d. n.d. n.d. 12 n.d. a0.01 a0.01 a0.01 a0.01 n.d. n.d. 13 n.d. 0.81 l 0.39 1.91 l 0.59 0.69 l 0.44 0.23 l 0.33 a0.01 n.d.

n.d.: not detectable. a) The given amounts are calculated as the mean of the individual urinary excretion determined for each of the eight volunteers by three independent clean-up and measurements.

N-[49-hydroxy-(E)-cinnamoyl]-L-aspartic acid (5), followed low amounts ranging between 0.3 and 1.9 lg (Table 2). In by N-[49-hydroxy-(E)-cinnamoyl]-L-glutamic acid (6) and comparison, N-[49-hydroxy-(E)-cinnamoyl]-L-tryptophane N-[49-hydroxy-(E)-cinnamoyl]-L-tyrosine (8) for which (12) was detectable only in trace amounts below the LOQ, amounts of 22.17 l 12.57 and 18.91 l 5.52 lg were found and N-[39,49-dihydroxy-(E)-cinnamoyl]-L-glutamic acid (2) (Table 2). N-[39,49-dihydroxy-(E)-cinnamoyl]-3-hydroxy-L- and N-[39,49-dihydroxy-(E)-cinnamoyl]-L-tryptophane (11) tyrosine (3), N-[39,49-dihydroxy-(E)-cinnamoyl]-L-tyrosine were not detectable at all. (4), N-[49-hydroxy-(E)-cinnamoyl]-L-dopa (7), N-cinna- Moreover, it is interesting to notice that five out of eight moyl-L-aspartic acid (10), and N-[49-hydroxy-39-methoxy- volunteers showed a significantly higher amount of amides (E)-cinnamoyl]-L-tyrosine (13) were present just in rather in the excreted urine (Fig. 4A), whereas the other three vol-

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com Mol. Nutr. Food Res. 2008, 52, 1201 – 1214 1209

Figure 4. Influence of individuals on the absolute amounts (lg/urine sample) of N-phenylpropenoyl-L-amino acids 1, 5, 6, 8, and 9 in urine excreted over a time period of 8 h. Amounts and SD are given as average values obtained from (A) five individuals excreting higher amounts of amides and (B) three individuals excreting lower amounts of amides.

Table 3. Total amount of N-phenylpropenoyl-L-amino acids determined in the administered cocoa drink, the total amount of urine collected over 8 h, and the recovery rate

Compound (no.) Amount (lg) ingesteda) Amount (lg) excretedb) Recovery (%)c)

N-[49-hydroxy-(E)-cinnamoyl]-L-glutamic acid (6) 77.7 44.6 57.3 N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]-L-tyrosine (13) 15.8 3.6 22.8 N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]-L-aspartic acid (9) 259.8 21.6 8.3 N-[49-hydroxy-(E)-cinnamoyl]-L-aspartic acid (5) 2003.4 149.9 7.5 N-[49-hydroxy-(E)-cinnamoyl]-L-tyrosine (8) 828.2 44.0 5.3 N-[cinnamoyl]-L-aspartic acid (10) 44.7 0.7 1.6 N-[49-hydroxy-(E)-cinnamoyl]-L-dopa (7) 110.7 1.1 1.0 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-dopa (3) 449.5 1.8 0.4 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-aspartic acid (1) 5063.6 15.0 0.3 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-tyrosine (4) 168.5 0.3 0.2 N-[39,49-dihydroxy-(E)-cinnamoyl]-L-glutamic acid (2) 67.6 n.d. n.c. N-[39,49-dihydroxy-(E)-cinnamoyl]-L-tryptophane (11) 8.1 n.d. n.c. N-[49-hydroxy-(E)-cinnamoyl]-L-tryptophane (12) 7.1 <0.1 n.c. n.d.: not detectable. n.c.: not calculated. a) Amount of amides in the administered cocoa drink is given as the mean of three independent clean-ups and measurements of the sample. b) Average amount of amide in total urine collected over 8 h for eight volunteers. c) The recovery rate was calculated in percent from the ingested amount and the excreted amount of amides. unteers excreted comparatively lower amounts (Fig. 4B), 149.9 lg in the total urine volume was found for N-[49- e.g., the group A excreted 76.83 lg of 5 in the total urine hydroxy-(E)-cinnamoyl]-L-aspartic acid (5), followed by collected after 2 h, whereas only 41.69 lg of that compound 44.6 and 44.0 mg found for N-[49-hydroxy-(E)-cinnamoyl]- was detectable in the urine sample taken form group B. L-glutamic acid (6) and N-[49-hydroxy-(E)-cinnamoyl]-L- To calculate the recovery of the individual N-phenylpro- tyrosine (8), respectively. By far the highest recovery rates penoyl-L-amino acids 1–13 in urine, that means the per- of 57.3 and 22.8% were found for N-[49-hydroxy-(E)-cinna- centage amount of ingested amides excreted with the urine, moyl]-L-glutamic acid (6) and N-[49-hydroxy-39-methoxy- the sum amount of each amide was calculated for the total (E)-cinnamoyl]-L-tyrosine (13), respectively. In compari- volume of urine collected over a time period of 8 h after son, only 8.3, 7.5, or 5.3% of the amount of N-[49-hydroxy- intervention. As given in Table 3, the highest amount of 39-methoxy-(E)-cinnamoyl]-L-aspartic acid (9), N-[49-

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com 1210 T. Stark et al. Mol. Nutr. Food Res. 2008, 52, 1201 – 1214

Figure 5. MS/MS (ESI – ) spectrum obtained for synthetic caffeoyl-3-O-b-D-glucuronide (15) and caffeoyl-4-O-b-D-glucuronide (16), respectively.

hydroxy-(E)-cinnamoyl]-L-aspartic acid (5), and N-[49- by acidic esterification of caffeic acid with methanol, was hydroxy-(E)-cinnamoyl]-L-tyrosine (8) present in the reacted with excess amounts of methyl acetobromo-a-D- ingested cocoa drink were excreted with the urine (Table 3). glucuronate in presence of silver trifluoromethane sulfonate In comparison, compounds 10 and 7 showed only a low at –208C (Fig. 2). Column chromatography on silica gel, recovery rate of 1.6 and 1.0%, respectively whereas the followed by de-protection using aqueous barium hydroxide, recovery rates of the remaining amides were found to be and purification by means of preparative RP-HPLC yielded below 0.4%. the target compounds caffeic acid 3-O-b-D-glucuronide In order to gain first insights into the metabolism of (15) and caffeic acid 4-O-b-D-glucuronide (16) in a ratio of N-phenylpropenoyl-L-amino acids in vivo, the urine samples l2:1. taken from two volunteers, one of the group A and the other Measurements of the exact masses of conjugates 15 and from group B (Fig. 4A and 4B), were treated with b-glucuro- 16 by means of high-resolution MS revealed the ion nidase/sulfatase at pH 5.0 for 4 h at 378C. Comparing the m/z 379.0637+ fitting well with the calculated mass of quantitative data obtained before and after enzyme treatment 379.0636 and confirming the sodium adduct + revealed exactly the same amounts of the N-phenylprope- (C15H16O10Na ) of the target compounds. LC-MS analysis noyl-L-amino acids in the corresponding urine samples, thus of 15 and 16 in the ESI – mode showed identical spectra indicating that neither O-glucuronides nor O-sulfates of the exhibiting m/z 355 [M – H] – as the pseudomolecular ion. polyphenolic amides were formed upon metabolic conjuga- Collision-induced fragmentation produced m/z 179 [M– tion. In order to confirm the functionality of the b-glucuroni- glucuronic acid–H] – as the most frequent daughter ion and dase/sulfatase enzyme solution used and to screen for caffeic m/z 135 as the second most frequent daughter ion (Fig. 5), acid O-glucuronides as potential metabolites of N-phenyl- thus indicating the release of a glucuronic acid moiety from propenoyl-L-amino acids, the two potential candidates caf- the molecule as well as the decarboxylation of the caffeic feic acid 3-O-b-D-glucuronide and caffeic acid 4-O-b-D-glu- acid moiety. curonide needed to be synthesized as reference compounds. The 1H NMR spectrum of compound 15 showed the five protons expected for the caffeic acid moiety. The doublets observed for H–C(7) and H–C(8) resonated at 7.42 and 3.2 Synthesis of caffeic acid O-b-D-glucuronides 6.84 ppm, respectively and exhibited a coupling constant of In order to synthesize both caffeic acid O-b-D-glucuro- 16 Hz, thus indicating a trans-configured double bond in nides, caffeic acid methylester (14), prepared in good yield the molecule. The protons H–C(2), H–C(5), and H–C(6)

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com Mol. Nutr. Food Res. 2008, 52, 1201 – 1214 1211

Figure 6. RP-HPLC-MS/MS (ESI – )-MRM analysis of (A) a reference solution of caffeoyl-O-b-D-glucuronides 15 and 16, and (B) a pooled urine sample collected 2 h after the intervention. were unequivocally assigned by means of homonuclear d,d- The isomer 16 showed similar NMR data as found for 1 correlation (g-COSY), and HMQC optimized for JC,H cou- compound 15, however, the hydroxyl group HO–C(49)was pling constants. The broad singlet detected at 9.30 ppm dis- significantly downfield-shifted to 5.78 ppm and the resid- appeared upon D2O addition and, therefore, was identified ual phenolic proton HO–C(3) was visible at 8.90 ppm as a as the residual aromatic hydroxyl proton. Carbon atoms broad singlet. Also the pattern of the aromatic protons was C(3) and C(4) could be clearly distinguished by means of slightly different. Proton H–C(6) resonating as double dou- 2 3 HMBC optimized for JC,H and JC,H (HMBC). Carbon atom blet at 7.09 ppm showed strong d,d-coupling to the neigh- C(3) showed an intense heteronuclear correlation to the H– boring proton H–C(5) and long-range coupling to H–C(2) 2 C(5) in the meta-position and a weak JC,H-coupling with in the COSYexperiment. Taking all spectroscopic data into H–C(2), whereas carbon C(4) coupled with the meta-pro- consideration, the structure of this compound was con- tons H–C(2) and H–C(6), respectively. As carbon C(3) firmed as caffeic acid 4-O-b-D-glucuronide (16, Fig. 5). showed also coupling with the anomeric proton H–C(19), compound 15 was identified as caffeoyl-3-O-b-D-glucuro- 3.3 Enzymatic hydrolysis of caffeic acid O-b-D- nide. The protons H–C(29) and H–C(39) of the glucuronic glucuronides acid moiety resonated at 3.30 ppm and showed strong d,d- couplings to H–C(49), which was slightly downfield shifted In order to confirm the functionality of the b-glucuroni- to 3.40 ppm and occurred as a pseudo triplet with a cou- dase/sulfatase enzyme solution, aqueous solutions of caf- pling constant of 9.2 Hz. Proton H–C(59) resonated at feic acid O-b-D-glucuronides 15 or 16, respectively were 3.98 ppm as a doublet and coupled with H–C(49) as con- treated individually with b-glucuronidase/sulfatase at firmed by a COSY experiment. Furthermore, the coupling pH 5.0 for 6 h at 378C and, after 0.5, 1, 2, 3, 4, and 6 h, ali- constant of 7.2 Hz observed for the anomeric proton H– quots were analyzed by means of HPLC-MS/MS. Already C(19) indicated the b-configuration of the glucuronide. The after 30 min, both caffeic acid O-b-D-glucuronides were hydroxyl groups HO–C(29) and HO–C(39) resonated as nearly completely hydrolyzed and after 1 h not even trace broad singlets at 5.22 and 5.47 ppm but could not be further amounts of the conjugates were detectable any more, thus differentiated, whereas proton HO–C(49) resonated as a confirming the functionality of the enzymes used (data not small, broadened signal at 4.49 ppm. The acidic protons shown). HOOC–(69) and HOOC–(9) were detectable as a very In order to screen for caffeic acid O-b-D-glucuronides as broad coalescent signal resonating at around 12.3 ppm. potential metabolites formed from N-phenylpropenoyl-L-

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com 1212 T. Stark et al. Mol. Nutr. Food Res. 2008, 52, 1201 – 1214

amino acids, urine samples collected were screened for 15 and N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]-L-aspartic and 16 by means of LC-MS/MS (Fig. 6). Independent of the acid (9), followed by N-[49-hydroxy-(E)-cinnamoyl]-L- urine sample collected, not even trace amounts of caffeoyl- aspartic acid (5) and N-[49-hydroxy-(E)-cinnamoyl]-L-tyro- O-b-D-glucuronides were detectable. These data clearly sine (8), all of which are bearing either a p-coumaroyl or a demonstrate that caffeic acid glucuronides are not the pre- feruloyl moiety in the molecule. It is interesting to notice, dominant metabolites of N-phenylpropenoyl-L-amino acids that all the N-[39,49-dihydroxy-(E)-cinnamoyl]-L-amino in humans. acids were found in much lower amounts in urine when compared to the amides bearing N-[49-hydroxy-39- methoxy-(E)-cinnamoyl]- and N-[49-hydroxy-(E)-cinna- 4 Discussion moyl]-moiety, respectively; for example, the recovery rates found for the caffeoyl derivatives 1–4, and 11 were found To date, there are no reports available on the absorption and to be below 0.4% (Table 3). bioavailability of N-phenylpropenoyl-L-amino acids, a Based on this finding, it might be concluded that after recently identified family of major polyphenols in T. cacao ingestion of N-[39,49-dihydroxy-(E)-cinnamoyl]-L-amino [9–12]. First, the amounts of the polyphenolic amides 1– acids, these N-caffeoyl-amides are metabolically conju- 13 (Fig. 1) were quantitatively determined in a sweetened gated to give the corresponding O-glucuronides, O-sulfates, cocoa drink by means of a recently developed SIDA [12]. and/or O-methyl ethers as reported in the literature for poly- The data, given in Table 1, revealed N-[39,49-dihydroxy-(E)- phenols including caffeic acid and caffeic acid esters such cinnamoyl]-L-aspartic acid (1) and N-[49-hydroxy-(E)-cin- as chlorogenic acid [24, 26–29]. Alternatively, it might be namoyl]-L-aspartic acid (5), followed by N-[49-hydroxy- argued that the N-[39,49-dihydroxy-(E)-cinnamoyl]-L-amino (E)-cinnamoyl]-L-tyrosine (8) and N-[39,49-dihydroxy-(E)- acids are either not efficiently absorbed, or are enzymati- cinnamoyl]-3-hydroxy-L-tyrosine (3) as the quantitatively cally hydrolyzed to release caffeic acid which, upon further predominating polyphenolic amides in the cocoa drink, conjugation, might be conjugated to give its corresponding thus being well in line with earlier data on these compounds O-glucuronide [27]. In order to check, if N-phenylprope- in roasted cocoa nibs [9, 12]. noyl-L-amino acids are metabolized in vivo to give their cor- After administration of a bolus amount of the cocoa drink responding O-glucuronide and/or O-sulfate conjugates, to eight healthy volunteers, the time-dependent urinary urine samples were comparatively analyzed by means of excretion of N-phenylpropenoyl-L-amino acid was quanti- LC-MS/MS-SIDA before and after treatment with b-glu- tatively determined by means of LC-MS/MS-SIDA. The curonidase/sulfatase. Surprisingly, the concentrations of the data obtained clearly demonstrated that independent from N-phenylpropenoyl-L-amino acids in the excreted urine was the structure of the N-phenylpropenoyl-L-amino acid, the not influenced by the enzymatic digestion, thus demonstrat- highest concentrations of these phytochemicals were ing that these polyphenolic amides are neither metabolized excreted with the urine already 2 h after consumption of the via their O-glucuronides nor their O-sulfates. cocoa drink (Table 2). Most surprisingly, the predominant To confirm the enzymatic functionality of the enzymes amide detected in urine was not the N-[39,49-dihydroxy-(E)- used for the digestion experiment, caffeic acid 3-O-b-D- cinnamoyl]-L-aspartic acid (1), which was found as the glucuronide (15) and caffeic acid 4-O-b-D-glucuronide (16) major N-phenylpropenoyl-L-amino acid in the cocoa drink were synthesized for the first time, and then treated with b- used in the intervention study (Table 2). In contrast, N-[49- glucuronidase/sulfatase. LC-MS/MS analysis of aqueous hydroxy-(E)-cinnamoyl]-L-aspartic acid (5), N-[49- solutions of caffeic acid O-b-D-glucuronides before and hydroxy-(E)-cinnamoyl]-L-glutamic acid (6), and N-[49- after enzymatic digestion clearly demonstrated a rapid hydroxy-(E)-cinnamoyl]-L-tyrosine (8) were found as the cleavage of the O-glucuronides to release caffeic acid and, quantitatively predominating amides in urine. Interestingly, in consequence, give strong evidence that N-phenylprope- five out of eight volunteers showed a significantly higher noyl-L-amino acids are not enzymatically hydrolyzed and amount of amides in the excreted urine (Fig. 4A), whereas metabolically conjugated as O-glucuronides or O-sulfates. the other three volunteers excreted comparatively lower In conclusion, this is the first human study demonstrating amounts (Fig. 4B). These findings give first evidence for that the N-phenylpropenoyl-L-amino acids 1, 3–10, 12, and major differences in the absorption of N-phenylpropenoyl- 13 are absorbed and excreted via the urine. LC-MS/MS L-amino acids and/or the metabolic activity of individuals. experiments before and after b-glucuronidase/sulfatase However, further studies with a higher sample number are treatment did not show any evidence that the N-phenylpro- needed to justify the formation of dichotomic groups. penoyl-L-amino acids are enzymatically hydrolyzed and Calculating the percentage amount of ingested N-phenyl- then conjugated with glucuronic acid as reported for chloro- propenoyl-L-amino acids excreted with the urine, by far the genic acids [27]. Bearing in mind that the N-[49-hydroxy-39- highest recovery rates of 57.3, 22.8, and 8.3% were found methoxy-(E)-cinnamoyl]- and N-[49-hydroxy-(E)-cinna- for N-[49-hydroxy-(E)-cinnamoyl]-L-glutamic acid (6), moyl]-L-amino acids were found in urine with rather high N-[49-hydroxy-39-methoxy-(E)-cinnamoyl]-L-tyrosine (13), recovery rates, it might be possible that these feruoyl- or

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com Mol. Nutr. Food Res. 2008, 52, 1201 – 1214 1213

p-coumaroyl-L-amino acids are formed by O-methylation [9] Stark, T., Bareuther, S., Hofmann, T., Molecular definition of or by reduction of the corresponding caffeoyl-L-amino the taste of roasted cocoa nibs (Theobroma cocoa) by means acids. As chocolate intake was recently shown to increase of quantitative studies and sensory experiments. J. Agric. Food Chem. 2006, 54, 5530–5539. the urinary excretion of m-hydroxyphenylpropionic acid, [10] Stark, T., Bareuther, S., Hofmann, T., Sensory-guided decom- ferulic acid, 3,4-dihydroxyphenylacetic acid, m-hydroxy- position of roasted cocoa nibs (Theobroma cocoa) and struc- phenylacetic acid, vanillic acid, and m-hydroxybenzoic acid ture determination of taste-active polyphenols. J. Agric. Food generated upon microbial metabolism of polyphenols [30], Chem. 2005, 53, 5407–5418. the intestinal microbial degradation of the N-phenylprope- [11] Stark, T., Hofmann, T., Isolation, structure determination, noyl-L-amino acids has to be taken into account for future synthesis, and sensory activity of N-phenylpropenoyl-L- studies. Quantitative studies on the levels of N-phenylpro- amino acids from cocoa (Theobroma cocoa). J. Agric. Food Chem. 2005, 53, 5419–5428. penoyl-L-amino acids as well as their putative metabolites in human plasma and urine are currently under investiga- [12] Stark, T., Justus, H., Hofmann, T., A stable isotope dilution analysis (SIDA) for the quantitative determination of N-phe- tion and will be published separately. nylpropenoyl-L-amino acids in coffee beverage and cocoa. J. Agric. Food Chem. 2006, 54, 2859–2867. We gratefully thank the International Graduate Research [13] Yoshihara, T., Yoshikawa, H., Sakamura, S., Sakuma, T., Clo- School Mnster, Germany/Nagoya, Japan (IRGT MS-NG vamides. L-Dopa conjugated with trans- and cis-caffeic acids 1143): “Complex Functional Systems in Chemistry: in red clover (Trifolium pratense). Agric. Biol. Chem. 1974, Design, Development, and Applications” funded by the 38, 1107–1109. Deutsche Forschungsgemeinschaft (DFG) and the Japan [14] Yoshihara, T., Yoshikawa, H., Kunimatsu, S., Sakamura, S., Society for the Promotion of Science (JSPS), as well as Sakuma, T., New amino acid derivatives conjugated with caffeic acid and DOPA from red clover (Trifolium pratense). PROvendis GmbH for supporting this work, and the volun- Agric. Biol. Chem. 1977, 41, 1679–1684. teers for participating in this study. [15] Tebayashi, S.-I., Ishihara, A., Tsuda, M., Iwamura, H., Induc- tion of clovamide by jasmonic acid in red clover. Phytochem- The authors have declared no conflict of interest. istry 2000, 54, 387–392. [16] Sanbongi, C., Osakabe, N., Natsume, M., Takizawa, T., et al., Antioxidative polyphenols isolated from Theobroma cacao. J. Agric. Food Chem. 1998, 46, 454–457. 5 References [17] Van Heerden, F. R., Brandt, E. V., Roux, D. G., Isolation and synthesis of trans- and cis-(–)-clovamides and their deoxy analogues from the bark of Dalbergia melanoxylon. Phyto- [1] Fisher, N. D., Hughes, M., Gerhard-Herman, M., Hollenberg, chemistry 1980, 19, 2125–2129. N. K., Flavan-3-ol-rich cocoa induces nitric-oxide-dependent vasodilation in healthy humans. J. Hypertens. 2003, 21, [18] Alemanno, L., Ramos, T., Gargadenec, A., Andary, C., Fer- 2281–2286. riere, N., Localization and idendification of phenolic compounds in Theobroma cacao L. somatic embryogenesis. Ann. [2] Janszky, I., Ericson, M., Blom, M., Georgiades, A., et al., Bot. 2003, 92, 613–623. Wine drinking is associated with increased heart rate variabil- ity in women with coronary heart disease. Heart 2005, 91, [19] Morishita, H., Takai, Y., Yamada, H., Fukuda, F., et al., Caf- 314–318. feoyltryptophan from green robusta coffee beans. Phyto- chemistry 1987, 26, 1195–1196. [3] Heiss, C., Dejam, A., Kleinbongard, P., Schewe, T., et al., Vascular effects of cocoa rich in flavan-3-ols. J. Am. Med. [20] Murata, M., Okada, H., Homma, S., Hydroxycinnamic acid Assoc. 2003, 290, 1030–1031. derivates and p-coumaroyl-(L)-tryptophan, a novel hydroxycinnamic acid derivate, from coffee beans. Biosci. Biotech- [4] Taubert, D., Berkels, R., Roesen, R., Klaus, W., Chocolate nol. Biochem. 1996, 59, 1887–1890. and blood pressure in elderly individuals with isolated sys- tolic hypertension. J. Am. Med. Assoc. 2003, 290, 1029– [21] Clifford, M. N., Kellard, B., Ah-Sing, E., Caffeoyltyrosine 1030. from green robusta coffee beans. Phytochemistry 1989, 28, 1989–1990. [5] Grassi, D., Lippi, C., Necozione, S., Desideri, G., Ferri, C., Short-term administration of dark chocolate is followed by a [22] Hensel, A., Deters, A. M., Mller, G., Stark, T., et al., Occur- significant increase in insulin sensitivity and a decrease in rence of N-phenylpropenoyl-L-amino acids in different blood pressure in healthy persons. Am. J. Clin. Nutr. 2005, 81, herbal drugs and influence on human keratinocytes, human 611–614. liver cells and against adhesion of Helicobacter pylori to human stomach. Planta Med. 2007, 73, 142–150. [6] Holt, R. R., Schramm, D. D., Keen, C. L., Lazarus, S. A., Schmitz, H. H., Chocolate consumption and platelet function. [23] Song, K.-S., Ishikawa, Y., Kobayashi, S., Sankawa, U., Ebi- J. Am. Med. Assoc. 2002, 287, 2212–2213. zuka, Y., N-Acylamino acids from Ephedra distachya cul- [7] Sies, H., Schewe, T., Heiss, C., Kelm, M., Cocoa polyphenols tures. Phytochemistry 1992, 31, 823–826. and inflammatory mediators. Am. J. Clin. Nutr. 2005, 81, [24] Scalbert, A., Williamson, G., Dietary intake and bioavailabil- 304S–312S. ity of polyphenols. J. Nutr. 2000, 13, 2073S–2085S. [8] Keen, C. L., Holt, R. R., Oteiza, P. I., Fraga, C. G., Schmitz, [25] Desmares, G., Lefebvre, D., Renevret, G., Le Drian, C., H. H., Cocoa antioxidants and cardiovascular health. Am. J. Selective formation of b-D-glucosides of hindered alcohols. Clin. Nutr. 2005, 81, 298S–303S. Helv. Chim. Acta 2001, 84, 880–889.

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com 1214 T. Stark et al. Mol. Nutr. Food Res. 2008, 52, 1201 – 1214

[26] Azuma, K., Ippoushi, K., Nakayama, M., Ito, H., et al., [29] Wenzel, E., Soldo, T., Erbersdobler, H., Somoza, V., Bioactiv- Absorption of chlorogenic acid and caffeic acid in rats after ity and metabolism of trans-resveratrol orally administered to oral administration. J. Agric. Food Chem. 2000, 48, 5496– Wistar rats. Mol. Nutr. Food Res. 2005, 49, 482–494. 5500. [30] Rios, L. Y., Gonthier, M.-P., Remesy, C., Mila, I. et al., Choc- [27] Gonthier, M.-P., Verny, M.-A., Besson, B., Rmsy, C., Scal- olate intake increases the urinary excretion of polyphenol- bert, A., Chlorogenic acid bioavailability largely depends on derived phenolic acids in healthy subjects. Am. J. Clin. Nutr. its metabolism by the gut microflora in rats. J. Nutr. 2003, 2003, 77, 912–918. 133, 1853–1859. [28] Zhu, B. T., Ezell, E. L., Lier, J. G., Catechol-O-methyltrans- ferase-catalyzed rapid O-methylation of mutagenic flavo- noids. J. Biol. Chem. 1994, 269, 292–299.

i 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mnf-journal.com